Simple robust estimation of gradient delays for spiral MRI

OBJECTIVE: Spiral MRI image quality is degraded by timing misalignments between gradient transmission and data sampling. Even well compensated and calibrated MRI scanners can experience residual, sequence specific gradient delays. Several methods exist for measuring and correcting for deviations in the k-space trajectory. This work proposes a simple, fast method for measuring gradient delays at several time points along a center-out spiral trajectory on any gradient channel. Using this method, independent delays can be calculated for all three gradient channels in six TR’s. METHODS: The proposed method plays a single spiral interleaf on the gradient channel of interest and then again with the gradient sign reversed. The resulting trajectories oscillate in opposite directions across the origin along the corresponding axis in k-space (figure 1). Gradient delays will shift the data from the two waveforms in opposite directions. The gradient delays can be estimated by 1) Separating the data into several segments (figures 1-2) centered in time around the expected k-space zero crossings. Each segment should use data that falls within the relatively linear region of the gradient waveform that is centered about the zero crossings, 2) Flipping the segments from one waveform about its zero crossings, 3) Correlating the segments between the two waveforms. The location of the peak correlation with respect to a zero crossing is proportional to twice the gradient delay associated with the gradient channel. The results from the correlations can be averaged to yield a single delay. Spiral data were acquired on a GE Signa Excite 3.0T scanner. Data corresponding to the proposed method were collected for the x, y, and z gradient channels. In-vivo 3D data of the head were obtained with 1mm isotropic resolution using a uni-axial Spiral Projection Imaging acquisition (figure 3). A 5mm axial slice of a phantom was also obtained using a 2D spiral acquisition (figure 4).